Intel, of course, has taken an extremely dim view of this practice and has announced that boards sold in this matter are the sole responsibility of the motherboard vendor to support, as well as throwing out some standard FUD about overclocking and instability. End users running Springdale in this accelerated mode are not, according to Intel, "receiving the benefits of 875P chipset-based platforms with PAT functionality".
This isn't exactly true. From a performance standpoint, end users certainly do seem to be receiving the benefits of PAT functionality, with 865-based boards performing equally to 875-based boards, or even running faster in certain cases. Further, while Intel is correct that overclocking components can lead to failure, it's a given that at least the more reputable manufacturers will test their boards a great deal before shipping them out with this hyper-acceleration mode available. Intel's claimed observation of "actual failure" has yet to be born out in real life by independent reviewers - until they are, this is just more smoke from Chipzilla's corner of the room.
No doubt Intel is concerned about end-user experience, and yes, the manufacturer wants to protect its reputation as a provider of extremely stable products, but let's not forget the financial aspect of this equation. According to our sources, an Intel Canterwood chipset currently sells for $55-$60, while a Springdale chipset can be as cheap as $20. Assuming these numbers are accurate, it's not hard to see where Intel's making the money when it comes to chipset sales.
Problem is, that doesn't do much for the motherboard manufacturers, especially in a depressed market where high-end boards aren't selling in great quantities. Faced with razor-thin margins in the lower market sectors and slow sales at the high-end, board manufacturers are looking for a way to boost their bottom lines while simultaneously increasing product shipments. For motherboard manufacturers the prospect of PAT-functional Springdale boards is a wish come true, allowing them to offer Canterwood performance at a much lower price, while trousering the difference in chipset costs.
No wonder Intel is taking such a firm stance against PAT-enabled 865 boards - it'll not only cost them a tremendous amount of money, it could raise some uncomfortable questions as well. If it turns out that Springdale boards can use PAT with no ill effects and the boards remain stable, users are going to want to know why exactly they're expected to pay triple the cost for a Canterwood board when a Springdale board is no different in terms of performance or stability.
Intel has never liked such questions, as evidenced by the fact that we're still waiting for solid information on what it is that makes the Xeon better for workstations and servers than the Pentium 4, but Santa Clara could find themselves facing more and more of them if PAT turns out to be an essentially free performance boost for the 865 chipset.
We'll be watching to see how this new breed of souped-up Springdale performs over the long term. Could be it's time for Intel to cut those chipset prices again. [Soon, Jack, soon. Ed.] µ
The 10 devices of Kicking PAT Gelsinger
How to hack Intel's PAT right off
Intel warns vendors not to overclock 865 chipsets
Intel names turbo mode chipset after Pat Gelsinger
Intel says there's no such thing as Kicking PAT
Intel's PAT creates hardware chaos
Kicking PAT Gelsinger kicked upstairs
Sign up for INQbot – a weekly roundup of the best from the INQ